Yes, I think that's exactly the insight that underlies this article, and that's why I think it makes sense. Unguided "exploration" coupled with large datasets leads wasting time on finding nonsense.

That doesn't mean flexibility in a tool isn't warranted - it's just that such flexibility should have a goal, and people should be discouraged from aimlessly applying it and then claiming they found something important.

(Also "exploration" is kind of a weasel word for student assignment; you can call your half-assed project "exploratory" and spin a good story from it that will give you a grade without any real work on your part.)